Arginine vasopressin (AVP) has been shown to have several non-renal actions including the potentiation of learned avoidance behavior in rats and improvement in cognitive functioning in humans. Research in our laboratory has confirmed these behavioral effects in rats using both peripheral and central injection of AVP. We have begun to examine the physiological basis for these effects. Peripheral administration of a vasopressor AVP antagonist reversed the prolongation of extinction produced by peripherally administered AVP in both active and passive avoidance, but also reversed the aversive unconditioned effects of AVP. However, central administration of the vasopressor AVP antagonist reversed peripheral effects of AVP only at doses shown to act peripherally to reverse vasopressor effects of AVP. An osmotic stress in doses known to liberate endogenous AVP mimicked the behavioral effects of exogenously administered AVP, and this stress effect was reversed by the AVP antagonist. These results support our hypothesis of separate but parallel AVP systems in the pituitary and brain with a role in behavioral adaptation to certain types of stress.